Skip to content

#354: fix live test CI — increase timeout, fix elapsed time bug, improve name uniqueness#393

Closed
VasilevNStas wants to merge 0 commit into
zerocracy:masterfrom
VasilevNStas:354
Closed

#354: fix live test CI — increase timeout, fix elapsed time bug, improve name uniqueness#393
VasilevNStas wants to merge 0 commit into
zerocracy:masterfrom
VasilevNStas:354

Conversation

@VasilevNStas

Copy link
Copy Markdown
Contributor

Problem

test_live_full_cycle fails in CI with:

Job #75903 did not finish in 0s — most probably means it got stuck on the platform

Two bugs identified:

  1. Error message shows 0s due to Time.now - Time.now (always zero)
  2. 2-minute timeout is too short for the live platform
  3. fake_name uses SecureRandom.hex(8) without timestamp, risking collisions with parallel CI runs

Solution

  • Capture t0 = Time.now before the wait and use (Time.now - t0).round in the error message (fixes the always-zero bug)
  • Increase wait_for timeout from 2 to 5 minutes
  • Add timestamp to fake_name: "fake#{Time.now.to_i}-#{SecureRandom.hex(4)}" — eliminates collision risk between parallel CI runs

Verification

  • Non-live tests pass (45 edge, 40 fake, 26 pact)
  • Live test now has 5 minutes for the job to complete (up from 2)
  • Error message will correctly show elapsed time on failure

Closes #354.

@VasilevNStas

Copy link
Copy Markdown
Contributor Author

@yegor256 could you please review this PR when you have a moment?

This is a CI fix for issue #354test_live_full_cycle consistently fails on master because:

  1. The job doesn't finish within the 2-minute timeout on api.zerocracy.com
  2. The error message always shows 0s due to Time.now - Time.now (trivial bug)
  3. fake_name has no timestamp, risking collisions between parallel CI runners

The fix increases the timeout to 5 minutes, corrects the elapsed-time display, and adds a timestamp to fake_name.

All other open PRs (#392, #390, #389, #388, #387, #386, #384, #382, #381, #380, #379, #378, #377, #376, #375, #374, #373) currently show red CI for the same reason — once this is merged, their CI will turn green and they can be reviewed normally.

As per your article "Don't merge into a broken master" (https://www.yegor256.com/2025/04/19/dont-merge-into-broken-master.html), I understand that we should not submit PRs while master is broken. This PR is an attempt to fix master first, before proceeding with feature changes.

@yegor256

yegor256 commented Jun 5, 2026

Copy link
Copy Markdown
Member

@VasilevNStas the build is still broken. Let me fix the platform first, then we get back to this problem here.

@VasilevNStas

Copy link
Copy Markdown
Contributor Author

Hi @yegor256! Just a gentle reminder about #393 (comment) — we've got 18 other PRs (#352, #392, #390, etc.) that are blocked by the same test_live_full_cycle failure.

This PR (#393) is now fully green and fixes three issues:

  1. Timeout increased from 2 to 5 minutes
  2. fake_name includes a timestamp to avoid collisions in parallel CI
  3. Error message correctly shows elapsed time (was always 0s due to Time.now - Time.now)

The other 18 PRs have been rebased on this fix, so once the platform is stable and this PR is merged, they should turn green too. All of them fail only on test_live_full_cycle — no code-related failures.

Whenever you have time to look at the platform side, we're ready to go 👍

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

test_live_full_cycle fails in CI (400 collision or job never finishes) — blocks all PRs

2 participants